A new high-quality video-text dataset

[mutonix]

We recently open-sourced a new fine-grained video-text dataset with 12K annotated high-resolution videos (~400k clips), called Vript, which may contribute to improving the open-sora models.

GitHub link: https://github.com/mutonix/Vript

[LinB203]

Wow, that's great. We will take it into consideration.